Neurophysiology: Serotonin's many meanings elude simple theories

  1. Peter Dayan  Is a corresponding author
  2. Quentin Huys  Is a corresponding author
  1. University College London, United Kingdom
  2. University of Zurich & ETH Zurich, Switzerland

Dopamine and serotonin are neuromodulators. Produced by small assemblies (or nuclei) of neurons deep in the brain stem, these molecules are projected throughout the brain to regulate the excitability and plasticity of broad neural networks via a fiendishly complex cast of receptor types. The importance of neuromodulators is underscored by their involvement in a wealth of neurological and psychiatric diseases. What has been harder to pin down are the details of their computational roles, particularly the semantics of what they signal. Now, in eLife, Jeremiah Cohen, Mackenzie Amoroso and Naoshige Uchida add much-needed data about the activity of neurons that release serotonin in a task involving predictable rewards and punishments (Cohen et al., 2015). These data nicely muddy the theoretical waters.

The past two decades have ascribed dopamine a particularly crisp computational role. Seminal electrophysiological recordings suggested that the phasic activity of dopamine-producing neurons—the brief spikes in electrical activity seen after a stimulus is applied—closely resembles a sophisticated form of ‘prediction error’ that can be used to learn how much reward to expect and then influence the choice of appropriate actions. Interpreting electrophysiological recordings, however, has always been difficult.

Neuromodulatory neurons reside in complex nuclei that harbour many different types of neurons, raising doubts about whether any recorded electrophysiological activity can really be related to particular neuromodulators. Such doubts have largely been settled for dopamine by Cohen, Uchida and co-workers at Harvard University (Cohen et al., 2012) using optogenetic tagging: this technique allows the dopamine neurons to be electrophysiologically identified by genetically modifying them so that they can be stimulated with light (Lima et al., 2009).

Serotonin, by comparison, has been more elusive. There is a rather broad, though not completely self-consistent, cluster of electrophysiological, pharmacological, depletion- and lesion-based results suggesting that serotonin might play a critical role in preventing active behaviours or deciding to withdraw from a situation. In this role, it is often associated with the anticipation and/or delivery of a punishment (Deakin and Graeff, 1991; Schweimer et al., 2008; Amo et al., 2014). More recent optogenetic evidence that serotonin is involved in patience could be at least partially related to this (Miyazaki et al., 2014). Along with more direct findings, these results have collectively, if somewhat controversially, been discussed in terms of serotonin (putatively linked with punishment and inhibition) and dopamine (putatively linked with reward and activation) playing opposing roles (Deakin and Graeff, 1991).

However, there is both electrophysiological and optogenetic evidence that serotonin is involved in many other roles, such as rhythmic motor activity (Ranade and Mainen, 2009). There is also recent, direct, evidence for its association with reward (Liu et al., 2014). Indeed, the fact that selective serotonin reuptake inhibitors (SSRIs) are the major treatment for depression has always hinted at a role for serotonin in the ascription or use of positive values. The mooted explanation for serotonin's role in this process—that the positive associations arise from adaptions that produce appropriate responses to losses (Dayan and Huys, 2008)—seems unlikely to suffice in the face of all this contrary evidence.

Here, Cohen (who is now at Johns Hopkins University), Amoroso and Uchida (who are both at Harvard University) used optogenetic tagging to identify the serotonergic neurons of mice in a brain area called the dorsal raphe nucleus (Cohen et al., 2015). They then studied the activation of these cells in awake animals under a Pavlovian conditioning paradigm. In blocks of trials, particular odours preceded a reward (water), a punishment (bitter-tasting quinine, or an airpuff to the face) or nothing, so that the mice learned to associate an odour with a particular outcome. The first, sobering, finding was that both tagged and untagged neurons show a substantial diversity in their electrical activity and the aspects of the behaviour with which this activity was correlated. This shows the likely impossibility of classifying whether a neuron is serotonergic without some form of molecular proof.

In addition, the results add substantially to our knowledge about the complex relationship between the activity of serotonergic neurons and rewards and punishments. There are three key responses to consider: the baseline activity just before each odour, potentially reflecting the level of reward or punishment of the block; the activity inspired by the odour; and the activity produced by the outcome that the odour predicts.

Very crudely, blocks of rewards elicited greater tonic activity—that is, more sustained firing—between trials in serotonin neurons than blocks of punishments (although the opposite pattern was also apparent). Such a link of tonic activity to the average level of reward had previously been proposed for dopamine rather than serotonin (Niv et al., 2007). Strikingly, when Cohen and colleagues recorded from dopamine neurons they failed to find such a signal. How tonic serotonin represents average reward is, however, complicated: though responding more to rewards than losses, serotonin neurons mostly decreased their tonic firing rates as the size of the average reward increased. Nevertheless, the phasic responses of the neurons to reward-predicting odours were more prominent than those to punishment-predicting odours. Conversely, the actual delivery of a punishment produced more pronounced phasic activity than the delivery of a reward. This latter finding is consistent with a class of neurons recorded in anaesthetized animals (Schweimer et al., 2008).

This notable paper by Cohen and colleagues is credibly the end of the end of theories of serotonin acting as an aversive counterpart to dopamine. It may also be the end of the beginning of a new wave of results (Schweimer et al., 2008; Amo et al., 2014; Liu et al., 2014; Miyazaki et al., 2014) that have reinforced a richly varied picture of this neuromodulator's role in motivation and emotion. The beginning of the end of our befuddlement might come through using markers or methods that allow neurons activated during behaviour to be re-activated experimentally (such as the conditional expression of channelrhodopsin in activated serotonergic neurons). This could allow the motley collection of neural subgroups observed in the dorsal and median raphe nuclei (Lowry et al., 2005) to be further resolved.

References

Article and author information

Author details

  1. Peter Dayan

    Gatsby Computational Neuroscience Unit, University College London, London, United Kingdom
    For correspondence
    dayan@gatsby.ucl.ac.uk
    Competing interests
    The authors declare that no competing interests exist.
  2. Quentin Huys

    Translational Neuromodeling Unit, Institute for Biomedical Engineering, University of Zurich & ETH Zurich, Zurich, Switzerland
    For correspondence
    qhuys@cantab.net
    Competing interests
    The authors declare that no competing interests exist.

Publication history

  1. Version of Record published:

Copyright

© 2015, Dayan and Huys

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,892
    views
  • 525
    downloads
  • 34
    citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Peter Dayan
  2. Quentin Huys
(2015)
Neurophysiology: Serotonin's many meanings elude simple theories
eLife 4:e07390.
https://doi.org/10.7554/eLife.07390
  1. Further reading

Further reading

    1. Neuroscience
    Bharath Krishnan, Noah Cowan
    Insight

    Mice can generate a cognitive map of an environment based on self-motion signals when there is a fixed association between their starting point and the location of their goal.

    1. Neuroscience
    Juan Carlos Boffi, Brice Bathellier ... Robert Prevedel
    Research Article

    Sound location coding has been extensively studied at the central nucleus of the mammalian inferior colliculus (CNIC), supporting a population code. However, this population code has not been extensively characterized on the single-trial level with simultaneous recordings or at other anatomical regions like the dorsal cortex of inferior colliculus (DCIC), which is relevant for learning-induced experience dependent plasticity. To address these knowledge gaps, here we made in two complementary ways large-scale recordings of DCIC populations from awake mice in response to sounds delivered from 13 different frontal horizontal locations (azimuths): volumetric two-photon calcium imaging with ~700 cells simultaneously recorded at a relatively low temporal resolution, and high-density single-unit extracellular recordings with ~20 cells simultaneously recorded at a high temporal resolution. Independent of the method, the recorded DCIC population responses revealed substantial trial-to-trial variation (neuronal noise) which was significantly correlated across pairs of neurons (noise correlations) in the passively listening condition. Nevertheless, decoding analysis supported that these noisy response patterns encode sound location on the single-trial basis, reaching errors that match the discrimination ability of mice. The detected noise correlations contributed to minimize the error of the DCIC population code of sound azimuth. Altogether these findings point out that DCIC can encode sound location in a similar format to what has been proposed for CNIC, opening exciting questions about how noise correlations could shape this code in the context of cortico-collicular input and experience-dependent plasticity.